whefs  Timeline

Many hyperlinks are disabled.
Use anonymous login to enable hyperlinks.

50 most recent check-ins

2009-06-16
23:56
backed out changes from commit [dd959311de]. They lead to a leak via whefs_string_copy_cstring(). Leaf check-in: [a4fb683398] user: stephan tags: trunk
2009-06-15
21:29
factored out whefs_inode::name. Had to add whefs_file::name, at least temporarily, so whefs_file_name_get() would still work. check-in: [dd959311de] user: stephan tags: trunk, forked_to_google_code
21:21
removed a debug message check-in: [e31d0d9e5b] user: stephan tags: trunk
20:57
added license header (dual Public Domain/MIT) in prep for eventual (possible) move to Google Code check-in: [e2162fef3b] user: stephan tags: trunk
20:32
fixed part of a small memleak when an fs is closed while inodes are still held open. To fix the second part (leaked whefs_fs or whio_dev objects) we need to track those in an is-opened list like we do inodes (but that's a bummer). check-in: [fce4e23351] user: stephan tags: trunk
20:28
removed some unused code. check-in: [c06f6055d2] user: stephan tags: trunk
20:28
added some internal docs. renamed a func. check-in: [af4c0c30da] user: stephan tags: trunk
20:11
minor C99 incompatibility fix for tcc. check-in: [e59a1c5a65] user: stephan tags: trunk
19:28
added compile-time option to disable inode names cache. Disabled by default because the cost is so high for many use cases. TODO: make it runtime-togglable. check-in: [48c63d9789] user: stephan tags: trunk
19:27
Added missing .o to test app check-in: [a49cd174e1] user: stephan tags: trunk
19:27
updated with 'doxygen -u' check-in: [a31f7f44c3] user: stephan tags: trunk
19:14
Fixed a bug in the handing of -D/--debug-flags when it was the last argument. Now no longer throws error when no FS filename is given when (openMode==WHEFSApp_NoOpen). check-in: [042766574f] user: stephan tags: trunk
19:07
fixed a minor hash list sorting issue where the list claimed to be unsorted when it had less than two members. check-in: [5c4a73ae54] user: stephan tags: trunk
18:06
minor doc correction check-in: [e8b0c6a4ef] user: stephan tags: trunk
2009-06-14
14:38
fixed a typo check-in: [52b0c9265b] user: stephan tags: trunk
13:49
corrected debug flags (removed in last commit) check-in: [9eaac3c869] user: stephan tags: trunk
13:49
added --debug-flags/-D to the command-line tools. check-in: [6dfd8af61b] user: stephan tags: trunk
12:40
minor restructuring check-in: [44e6033507] user: stephan tags: trunk
12:39
added flags arg to whefs_setup_debug() check-in: [0936187755] user: stephan tags: trunk
12:27
added some debug output showing the amount of mem used by the name strings cache. check-in: [a3a5437a34] user: stephan tags: trunk
12:26
removed whefs_string_db::buf member. check-in: [23f72ed5d1] user: stephan tags: trunk
12:09
minor refactoring check-in: [0ccbdbf787] user: stephan tags: trunk
12:09
minor refactoring. Fixed an off-by-one in whefs_inode_hash_cache_load(). Added whefs_string_db_memcost() check-in: [c305b47727] user: stephan tags: trunk
12:07
added #warning when WHEFS_FS_BITSET_CACHE_ENABLED==0 check-in: [ce33934247] user: stephan tags: trunk
12:07
refactored whefs_inode_by_name() significantly. check-in: [b145aa760f] user: stephan tags: trunk
12:01
minor pedantic memset() check-in: [d18c8042e6] user: stephan tags: trunk
2009-06-13
19:38
minor doc corrections check-in: [06c30ae9a4] user: stephan tags: trunk
06:36
integrated a new name caching mechanism which stores the name strings. Has a non-trivial (but not huge) memory cost, but we can now search for an inode ID by name in close to O(1) time (best case) once an inode's name has been traversed once. check-in: [8681a20356] user: stephan tags: trunk
04:23
comitting a bit of a hodgepodge which seems to work. In the middle of adding the second name cache. check-in: [0117245e52] user: stephan tags: trunk
03:13
removed an extraneous flush() check-in: [7d0ee96650] user: stephan tags: trunk
03:11
doc updates only check-in: [1d384b9273] user: stephan tags: trunk
02:44
whefs_string_clear() now does not free(string) if self->alloced==0, to allow strings to be shallowly copied. check-in: [9b324a7a86] user: stephan tags: trunk
02:40
Added whio_dev_ioctl_BUFFER_uchar_ptr for whio_dev_membuf. Slightly changed semantics of whio_dev_membuf expansion factor==1.0. s/whio_blockdev_id/whio_size_t/g check-in: [fcfcee82e6] user: stephan tags: trunk
2009-06-12
20:34
s|inode_name_cache|inode_hash_cache|g check-in: [c6328e0f93] user: stephan tags: trunk
20:26
refactored some of the more pedantic consistency checks (which are called very often, ~1% of the runtime in one of my tests) so that they can be compiled as macros or functions, allowing us to eventually assert() then or compile then out if they end up being too pedantic. check-in: [d3c2f745a6] user: stephan tags: trunk
20:22
s/VFS/EFS/g check-in: [0b80b4a58e] user: stephan tags: trunk
20:21
minor doc updates check-in: [a62c1b065b] user: stephan tags: trunk
20:20
minor portability fixes for an older gcc. check-in: [a2cee0cc32] user: stephan tags: trunk
20:20
minor portability fixes for an older gcc. check-in: [3bb7af01aa] user: stephan tags: trunk
20:17
s/VFS/EFS/g check-in: [b43594ea76] user: stephan tags: trunk
2009-06-11
18:02
Fixed erroneous warning message when trying to read from right at EOF, which happened often when the file size is a multiple of the block size and we read in block-sized chunks. check-in: [7e35fa689e] user: stephan tags: trunk
15:18
included the whio_dev object alloc pool size for some testing. check-in: [ad5ae78c61] user: stephan tags: trunk
15:17
nodedev now only flushes if it is in r/w mode check-in: [2aa7e45d47] user: stephan tags: trunk
14:18
added an initial name lookup cache. It's extremely basic and low-level, but it saves some i/o in cases where the same filename is sought after more than once (potentially inode_count-2 name seeks and reads saved per search-by-name). check-in: [c8ef23e825] user: stephan tags: trunk
13:50
now reduced the default block size if all files are smaller than that. check-in: [17803718e3] user: stephan tags: trunk
03:27
whefs_inode_name_get() now caches the hashcode for each filename, but the cache is not yet used anywhere. check-in: [f53f1c1eb3] user: stephan tags: trunk
2009-06-10
23:38
re-added whefs_hash.{c,h} cuz i think i've found a good way to use it for name caching. check-in: [e03e37686b] user: stephan tags: trunk
23:15
whefs_inode_foreach() now passed a const inode instead of a non-const one. check-in: [e3466ce49e] user: stephan tags: trunk
23:13
added whefs_inode_foreach(). Reimplemented whefs-ls on top of it. check-in: [3224e003d2] user: stephan tags: trunk
23:12
minor header reorgs check-in: [9052062853] user: stephan tags: trunk